At the Nashua Cloud .NET User Group meetup and Dev Boston User Group, I delivered a session titled “Building Enterprise Agents using Azure AI Search,” focusing on how modern enterprises can go beyond basic RAG and build knowledgeable, secure, multi-source AI agents. The talk centered on the evolution from simple search pipelines to agentic retrieval, granular security, multimodal ingestion, and enterprise-grade governance—all powered by Azure AI Search. AISearch
We began with the fundamentals: the query pipeline and data pipeline that make enterprise retrieval possible. Azure AI Search serves as the backbone, enabling vector search, semantic ranking, and the ingestion of multimodal data—from PDFs and documents to images, tables, and video. Logic Apps add automation and connectors for real-world ingestion workflows, helping organizations unify content across SharePoint, OneDrive, Azure Storage, and more. AISearch
The session also explored the shift from classic RAG to Agentic Retrieval—a new pattern where an LLM plans, decomposes, and orchestrates multiple search queries in parallel. Instead of a single query and a single answer, the model executes subqueries, merges results, re-ranks them, and uses grounded context to generate precise responses. This improves outcomes for complex enterprise queries such as policy lookups, compliance summaries, and cross-department knowledge requests. AISearch
The key takeaway:
Enterprise agents require more than RAG—they need structured retrieval, grounded reasoning, multimodal awareness, and strong access control. Azure AI Search provides the infrastructure to make that possible.
The Recording and sample codes are available here:
Watch the session:https://youtu.be/U77ufaeYiSU
Download Presentation: https://www.slideshare.net/slideshow/building-enterprise-agents-with-azure-ai-search-from-rag-to-agentic-retrieval/283097996
Sample Code: https://github.com/nhcloud/aisearch
If you missed the session, join us at the next Nashua Cloud .NET User Group (NashuaUG) or Dev Boston User Group meetup to continue exploring practical, real-world AI engineering.